연구보고서
- 저자
- 이승수 박사
- 작성일
- 2017.07.04
- 조회
- 224
- 요약
- 목차
In recent years, we are suffering from abnormal weather conditions due to global climate change. The abnormal weather conditions intensify extreme rainfall events, so that urban flooding and inundation caused by extreme rainfall events are occurring frequently. Once urban flooding and inundation take place, the damages are very huge and severe because there is a concentration of properties and populations in urbanized areas. Therefore, the development of countermeasures for flood defence is urgent. These potential countermeasures can be divided into two types: (1) structured, and (2) non-structured.
Structured countermeasures is defined as construction of hydraulic structures such as levee reinforcement, increase in sewerage capacities, installation of detention ponds, and flood control dams. Non-structured countermeasures involve the optimization of a dam operation system, preparing of a flood hazard map, and the development of a flood warning system and evacuation map. Structured countermeasures, although very clear in its impacts, require large amounts of funding and time to build. On the contrary, the impacts of non-structured countermeasures are difficult to quantify, but can minimize the damages caused by flooding and inundation with a low budget if we are able to predict damages with an acceptable level of accuracy.
This study develops a real-time urban inundation prediction system utilizing predicted precipitation data based on remote sensing technology. In order to select an appropriate numerical technique to simulate 2 dimensional (2D) surface flow, numerical methods are investigated. The cartesian coordination system model, general curvature coordinate model polygon grid model, and grid mesh refinement model are selected and investigated to be appropriate models for the study. In addition, the 1 dimensional (1D) sewerage system analysis model which was introduced by Lee et al. (2015) is used to simulate inlet and overflow phenomena by interacting with surface flow. The cartesian coordination system model and grid mesh refinement model are selected and combined with the 1D sewerage analysis model. Also parallel computing method, OpenMP, is applied to reduce the calculation time. The selected models were testede against the 25 August 2014 extreme rainfall event, which caused severe inundation damages in the Busan area in Korea. The Oncheoncheon basin is selected as the study basin and the observed radar data are assumed to be the predicted rainfall data. A 10m×10m resolution grid is used for the cartesian coordination model and a 10m×10m resolution grid mixed with a 5m×5m resolution grid is used for the mesh refinement model. A 5m×5m grid was used to depict road network in urban areas. The total grid number of the cartesian coordination model is 556,009, and is 899,575 for the mesh refinement model.
The High Performance Computer (HPC), located at the APEC Climate Center (APCC) in Busan, is used to conduct numerical simulations. Intel Xeon X5690 CPU is equipped, and the clock count is 3.46GHz, and the 12 cores are equipped into 1 node. Calculation speed was estimated by increasing the number of cores and finally, 12 cores was determined to be the best number to optimize the calculation speed.
Approximately 14 minutes of calculation are required to simulate a 3 hour prediction using the cartesian coordination model, and meaning that the model is able to secure a 166 minute lead time. Also, it takes 36 minutes of calculation to complete a 3 hour prediction using the mesh refinement model, meaning that the model is able to secure a 144 minute lead time.
When validating the models using observed data, each simulation result shows acceptable maximum inundation depth with propagation phenomena. In addition, the mesh refinement model shows more reasonable inlet effect into storm drain in mountainous areas, and an overflow effect onto surfaces in urbanized areas due to the mesh refinement models ability to depict the distribution of storm drain on the road. On the other hand, the cartesian coordination model is able to secure a longer lead time than the mesh refinement model.
Through this research, 1D-2D coupled urban inundation models are developed for real time inundation forecasting using remote sensing techniques. Both models show acceptable calculation speed with accuracy. Therefore, it is expected that both models can be used for the real time urban inundation forecasting, which has the potential to minimize damages.